eleg 5491: introduction to deep learning - python programming...
Post on 01-Apr-2021
31 Views
Preview:
TRANSCRIPT
ELEG 5491: Introduction to Deep LearningPython Programming Basics
Prof. LI Hongsheng, TA. Peng Gao
Office: SHB 428e-mail: hsli@ee.cuhk.edu.hk, 1155102382@link.cuhk.edu.hk
web: https://blackboard.cuhk.edu.hk
Department of Electronic EngineeringThe Chinese University of Hong Kong
Jan. 2021
Prof. LI Hongsheng, TA. Peng Gao ELEG 5491: Introduction to Deep Learning
Python
Open source general-purpose language
Object Oriented, Procedural, Functional
Easy to interface with C/ObjC/Java/Fortran
Easy-ish to interface with C++ (via SWIG)
Great interactive environment
Does not need compile;
Many good packages for various purposes: web, file, multi-thread, AI, etc.
Documentation: http://www.python.org/doc/
Free book: http://www.diveintopython.org
IMPORTANT: Learn to use Google. At most times, it will bring you tosome answers on Stack Overflow
Prof. LI Hongsheng, TA. Peng Gao ELEG 5491: Introduction to Deep Learning
Python version and distributions
Latest version is 3.7Version 2.7.x will not be supported after 2020
If you start from scratch, you should use Python 3.x
Python has different distributions that pre-install a large number of usefulpackages: numpy, scipy, scikit-learn, matplotlib, etc.For scientific computing, Anaconda is recommended
Download: https://www.anaconda.com/download/
Search package in Anaconda package poolsconda search scipy
Install package from Anaconda package poolsconda install scipy
One can also use the Python Package Index (PyPI) for packagemanagementpip install scipy
Prof. LI Hongsheng, TA. Peng Gao ELEG 5491: Introduction to Deep Learning
Python interactive interface
After installation, in your commnd window, type in “python”
Python has an interactive interface
% python Python 3.6.0b2+ (3.6:84a3c5003510+, Oct 26 2016, 02:33:55)[GCC 6.2.0 20161005]Type “help”, “copyright”, “credits” or “license” for more information.>>>
Python interpreter evaluates inputs
>>> 3*(7+2)27
Python prompts with “>>>”
To exit Python:
Ctrl-D
Prof. LI Hongsheng, TA. Peng Gao ELEG 5491: Introduction to Deep Learning
Run Python program via commands
A sample code
Sample code
def main():
x = 34 - 23 # A comment.
y = "Hello" # Another one.
z = 3.45
if z == 3.45 or y == "Hello":
x = x + 1
y = y + " World" # String concat.
print (x)
print (y)
if name = " main ":
main()
Execution results
Running results
12
Hello World
Prof. LI Hongsheng, TA. Peng Gao ELEG 5491: Introduction to Deep Learning
Run Python program via commands
Write a Python program using text editor and save it as “myscript.py”
Run Python program via command window
Run Python program
python myscript.py arg1 arg2
Prof. LI Hongsheng, TA. Peng Gao ELEG 5491: Introduction to Deep Learning
Syntax basics
Assignment uses = and comparison uses ==
For numbers, + - * / % are as expected
Logical operators are words (and, or, not), not symbols
The basic printing command is print
The first assignment to a variable creates itVariable types don’t need to be declared
Python figures out the variable types on its own
One cannot access non-existent names
Avoid using system pre-defined names
WhitespaceWhitespace is meaningful in Python: especially indentation and placementof newlines.
By default, four whitespace (” ”) = one indentation
Use a newline to end a line of code. Use \ when must to the next lineprematurally
No braces { } to mark blocks of code in Python. Use consistent indentationinstead
The first line with less indentation is outside of the block
The first line with more indentation starts a nested block
Prof. LI Hongsheng, TA. Peng Gao ELEG 5491: Introduction to Deep Learning
Syntax basics
CommentsStart comments with #: the rest of line is ignored
Can include a ”documentation string” as the first line of any new functionor class that you define
It is of good practice to include one to explain the function
def my function(x,y):"""This is the docstring.This function does blah blah blah."""# The code goes here
Basic datatypesIntegers (default for numbers)
z = 5 / 2 # Answer is 2, integer division
Floats
z = 3.456
StringCan use ”” or ’ ’ to specify.
"abc" ’abs’
Unmatched can occur within the string
"matt’s"
Use triple double-quotes for multi-line strings or strings than contain both ’ and” inside of them
"""a’b"c"""
Prof. LI Hongsheng, TA. Peng Gao ELEG 5491: Introduction to Deep Learning
Syntax basics
Naming rulesNames are case sensitive and cannot start with a number
They can contain letters, numbers, and underscores
bob Bob bob 2 bob bob 2 BoB
Reserved namesand, assert, break, class, continue, def, del, elif,
else, except, exec, finally, for, from, global, if,
import, in, is, lambda, not, or, pass, print, raise,
return, try, while
Prof. LI Hongsheng, TA. Peng Gao ELEG 5491: Introduction to Deep Learning
Control of flow
The following code are self-explanatory
Prof. LI Hongsheng, TA. Peng Gao ELEG 5491: Introduction to Deep Learning
Sequence types
TupleA simple immutable ordered sequence of itemsItems can be of mixed types, including collection types
StringsImmutableConceptually very much like a tuple
ListMutable ordered sequence of items of mixed types
All three sequence types (tuples, strings, and lists) share much of the samesyntax and functionality
Key difference: Lists are mutable
The operations shown can be applied to all sequence types
Prof. LI Hongsheng, TA. Peng Gao ELEG 5491: Introduction to Deep Learning
Sequence types
Tuples are defined using parentheses (and commas)
Lists are defined using square brackets (and commas)
Strings are defined using quotes (”, ’, or ”””).
Can access individual members using square bracket “array” notationIndices are 0-based
Prof. LI Hongsheng, TA. Peng Gao ELEG 5491: Introduction to Deep Learning
Indices & slicing
Positive index: count from the left, starting with 0
Negative lookup: count from right, starting with –1
Slicing: return a copy of the container with a subset of the originalmembers. Start copying at the first index, and stop copying before thesecond index
Can also use negative indices when slicing
Omit the first index to make a copy starting from the beginning of thecontainer
Omit the second index to make a copy starting at the first index and goingto the end of the container
Prof. LI Hongsheng, TA. Peng Gao ELEG 5491: Introduction to Deep Learning
Operators
in: test whether a value in a container or a substring in a string
+: produces a new tuple, list, or string whose value is the concatenation ofits arguments
Prof. LI Hongsheng, TA. Peng Gao ELEG 5491: Introduction to Deep Learning
Operators & assignment
*: produces a new tuple, list, or string that ”repeats” the original content.
For simple built-in datatypes (integers, floats, strings), assignment behavesas you would expectFor other mutable datatypes (lists, dictionaries, user-defined types),assignment works differently
When we change these data, we do it in placeWe don’t copy them into a new memory address each timeIf we type y=x and then modify y, both x and y are changed
Prof. LI Hongsheng, TA. Peng Gao ELEG 5491: Introduction to Deep Learning
Operations for lists
Lists are mutable
We can change lists in place
Name li still points to the same memory reference when we are done
The mutability of lists means that they are not as fast as tuples
append & insert
Prof. LI Hongsheng, TA. Peng Gao ELEG 5491: Introduction to Deep Learning
Operations for lists: extend and +
+ creates a fresh list (with a new memory reference)
extend operates on list li in place
Differences:
extend takes a list as an argument
append takes a singleton as an argument
Prof. LI Hongsheng, TA. Peng Gao ELEG 5491: Introduction to Deep Learning
Other operations for lists
IMPORTANT: Learn to read documentationhttps://docs.python.org/3/tutorial/datastructures.html
Prof. LI Hongsheng, TA. Peng Gao ELEG 5491: Introduction to Deep Learning
String formatting
Since Python 3.x, str.format() is recommended for formatting strings
With str.format(), the replacement fields are marked by curly braces
>>> name = ‘Eric’
>>> age = 74
>>> "Hello, {}. You are {}.".format(name, age)
‘Hello, Eric. You are 74.’
Reference variables in any order by referencing their index
>>> "Hello, {1}. You are {0}.".format(age, name)
‘Hello, Eric. You are 74.’
Reference variable names
>>> "Hello, {name}. You are {age}.".format(name=’Eric’,age=74)
‘Hello, Eric. You are 74.’
Specify detailed number format
>>> "Hello, {name}. You are {age:+.2f}.".format(name=‘Eric’, age=74)
‘Hello, Eric. You are +74.00.’
Prof. LI Hongsheng, TA. Peng Gao ELEG 5491: Introduction to Deep Learning
String formatting
A reference table for string formatting
Prof. LI Hongsheng, TA. Peng Gao ELEG 5491: Introduction to Deep Learning
Dictionaries
Dictionaries store a mapping between a set of keys and a set of values
Keys can be any immutable type
Values can be any type
A single dictionary can store values of different types
Can define, modify, view, lookup, and delete the key-value pairs in thedictionary
Prof. LI Hongsheng, TA. Peng Gao ELEG 5491: Introduction to Deep Learning
For loop
The for loop in python generally loop over a sequence or a dictionary
fruits = ["apple", "banana", "cherry"]
for x in fruits:
print(x)
range() function returns a sequence of numbers, starting from 0 bydefault, and increments by 1 (by default), and ends at a specified number.
for x in range(6):
print(x)
Increment the sequence from 2 to 30 (but not including 20) with 3(default is 1)
for x in range(2, 30, 3):
print(x)
Use enumerate() to output the numeric index in the loop
my list = [’apple’, ’banana’, ’grapes’, ’pear’]
for c, value in enumerate(my list):
print(c, value)
# Outputs:
# 1 apple
# 2 banana
Prof. LI Hongsheng, TA. Peng Gao ELEG 5491: Introduction to Deep Learning
Functions
def creates a function and assigns it a name. return sends a result backto the caller
Arguments are passed by assignment
Arguments and return types are not declared
Arguments are passed by assignment
Passed arguments are assigned to local names
Assignment to argument names don’t affect the caller
Changing a mutable argument may affect the caller
Prof. LI Hongsheng, TA. Peng Gao ELEG 5491: Introduction to Deep Learning
Functions
Can define default values for arguments that need not be passed
All functions in Python have a return value (even no return in the code)
Functions without a return return the special value None
There is no function overloading (functions can’t have the same name) inPython
Functions can be used as any other datatypes. They can beArguments to function
Return values of functions
Assigned to variables
Parts of tuples, lists, etc
Prof. LI Hongsheng, TA. Peng Gao ELEG 5491: Introduction to Deep Learning
Modules
Code reuse:routines can be called multiple times within a programRoutines can be used from multiple programs
Namespace partitioninggroup data together with functions used for that data
Implementing shared services or dataCan provide global data structure that is accessed by multiple subprograms
Modules are functions and variables defined in separate files
Items are imported using from or import
from module import function
function()
import module
module.function()
import module as md
md.function()
Modules are namespaces. Can be used to organize variable namesatom.position = atom.position - molecule.position
Prof. LI Hongsheng, TA. Peng Gao ELEG 5491: Introduction to Deep Learning
Classes & objects
A software item that contains variables and methods
Object Oriented Design focuses onEncapsulation: dividing the code into a public interface, and a privateimplementation of that interface
Polymorphism: the ability to overload standard operators so that they haveappropriate behavior based on their context
Inheritance: the ability to create subclasses that contain specializations oftheir parents
Example
Prof. LI Hongsheng, TA. Peng Gao ELEG 5491: Introduction to Deep Learning
Atom & molecule classes
Overloaded the default constructor
Object Oriented Design focuses on
Defined class variables (atno,position) that are persistent and local to theatom object
Good way to manage shared memory
Overloaded the print operator
Use the atom class to build molecules
Prof. LI Hongsheng, TA. Peng Gao ELEG 5491: Introduction to Deep Learning
top related